当前位置:网站首页>Design specification for smart speakers v1.0
Design specification for smart speakers v1.0
2022-06-30 08:44:00 【Intelligent speaker design】
Code for acoustic design of intelligent speakers V1.0
Preface
This specification is based on the author's current work experience , If there is any deficiency , Welcome to correct .
1. Microphone system design
1.1 MIC Model selection
- Smart speakers are right MIC High array consistency requirements , Due to the production process , simulation MIC Poor consistency , Therefore, numbers are preferred MIC;
- simulation MIC Its advantage is high signal-to-noise ratio , And the gain is adjustable , Under the condition of ensuring consistency , simulation MIC Have more advantages ;
- Sensitivity requirements : Greater than [email protected] 1kHz;
- Signal to noise ratio requirements : Greater than 64dB;
- Acoustic overload point (AOP) requirement : Greater than 120dBSPL
- MIC Monomer distortion requirements : Less than 1%@1kHz
1.2 MIC Sound outlet design
- The microphone hole on the housing should be larger than 0.8mm;
- The smaller the volume of the front cavity, the better , The shorter the hole length on the shell, the better , The larger the diameter of the hole, the better , Generally, the diameter depth ratio of the sound outlet is not less than 1:3;
- It can be completed with independent integrated structural parts MIC seal up & Construction of radio channel , Do not use multiple assemblies , The error chain in production will lead to the decrease of sealing consistency ;
- The dust screen can be selected with appropriate density and sound resistance , Choosing a dust net will increase the cost ;
- Bottom sound MIC The sealing structure is easy to do ;
- MIC The installation position shall be as far away from the noise source as possible ( Such as motor 、 Fans, etc );
- The sound outlet shall avoid the occurrence of the resonant cavity ( Small at both ends , Big in the middle );
1.3 MIC Formation selection
- MIC The more the number of , The better the result. , But the higher the cost ;
- MIC The formation is divided into ring array and linear array , The ring array shall be placed at equal intervals as far as possible MIC, Array diameter range 60~80mm, It is widely used in the industry 70mm;
- For example, the structure of unrestricted priority ring array . Affected by the array beam , The angle identification deviation of linear array in the axis direction is large .
2. Loudspeaker system design
2.1 Speaker unit selection
- Distortion of loudspeaker unit (THD) As small as possible , The less distortion , Echo cancellation (AEC) The better the result. , The higher the interrupt wake-up rate ;
- When the power is satisfied , Ferrite speakers are preferred , The cost is low , Ferrite also acts as a counterweight ;
- The maximum volume of the system shall not exceed the rated power of the loudspeaker unit ;
2.2 Loudspeaker structure design
- If the structure permits , The speaker and microphone should be as far away as possible ;
- The loudspeaker and microphone shall be in different cavities , Seal well , Prevent internal transmission ;
- The loudspeaker system shall be subject to shock absorption treatment , Prevent structural vibration from causing structural sound transmission , And keep a certain distance from other components , Prevent abnormal signal caused by collision ;
3. Circuit system design
3.1 Mining circuit design
- According to different mining positions, it can be divided into internal mining and external mining , The external recovery is divided into the recovery before the external power amplifier and the recovery after the external power amplifier , The recommended mining type is mining after external power amplifier , That is, the recovery is located between the power amplifier and the loudspeaker ( After the power amplifier , Before the loudspeaker , such AEC The best effect );
- If there is automatic gain control in the recovery circuit (AGC), It should be placed before the stoping position ;
3.2 Signal quality requirements
- Power matching : The device plays the full sweep signal at its maximum volume ,MIC The signal cannot be truncated ;
- Self broadcast and self recorded audio signal ,MIC The signal shall be later than the recovery signal , It cannot be earlier than the recovery signal ;
- Requirements for signal-to-noise ratio of the whole machine : Not less than [email protected] 1kHz;
- Sensitivity requirements of the whole machine : Not less than [email protected] 1kHz;
- MIC Too small signal will cause the far-field wake-up rate to decrease , When the device plays a full sweep signal at its maximum volume , The maximum amplitude range is -3dBFS~-9dBFS It is advisable to ;
边栏推荐
猜你喜欢
涂鸦Wi-Fi&BLE SoC开发幻彩灯带
Redis design and Implementation (VII) | publish & subscribe
el-input 限制只能输数字
微信公众号第三方平台开发,零基础入门。想学我教你啊
电流探头电路分析
【NVMe2.0b 14-3】Doorbell Buffer Config command、Device Self-test command
Advanced technology management -- how managers design and build echelons
Detailed explanation of pytoch's scatter function
【NVMe2.0b 14-1】Abort、Asynchronous Event Request、Capacity Management command
codeforces每日5题(均1700)-第三天
随机推荐
Redis design and Implementation (VI) | cluster (sharding)
Swagger use
About Lombok's @data annotation
Occasionally, Flink data is overstocked, resulting in checkpoint failure
Redis design and Implementation (V) | sentinel sentry
Gilbert Strang's course notes on linear algebra - Lesson 4
[data analysis and display]
快应用中实现自定义抽屉组件
Wechat official account third-party platform development, zero foundation entry. I want to teach you
Alcohol tester scheme: what principle does the alcohol tester measure alcohol solubility based on?
Flink Exception -- No ExecutorFactory found to execute the application
Interpretation of orientedrcnn papers
Vite project require syntax compatibility problem solving require is not defined
关于Lombok的@Data注解
el-input 限制只能输数字
TiDB 6.0:让 TSO 更高效丨TiDB Book Rush
国债逆回购绝对安全吗 网上怎么开户
layer.open 当传值为数组或值太长时处理方法
vite项目require语法兼容问题解决require is not defined
Detectron2 source code reading 3-- encapsulating dataset with mapper